Natural Cyclopeptide RA-V Inhibits the NF-ÎºB Signaling Pathway By

Wang et al. Cell Death and Disease (2018) 9:715 DOI 10.1038/s41419-018-0743-2 Cell Death & Disease

ARTICLE Open Access Natural cyclopeptide RA-V inhibits the NF- κB signaling pathway by targeting TAK1 Zhe Wang 1,2,SimengZhao2,LihuaSong1,YuzhiPu3, Qiang Wang4,GuangzhiZeng2, Xing Liu4,MingBai5, Shao Li 5,FabaoGao3, Lijuan Chen3, Chen Wang1,4 and Ninghua Tan 1,2

Abstract Rubiaceae-type cyclopeptides (RAs) are a type of plant cyclopeptides from the Rubia that have garnered significant attention owing to their unique bicyclic structures and amazing antitumour activities. Our recent work has shown that RAs suppress inflammation and angiogenesis and induce apoptosis. However, the underlying mechanism and targets remained unknown. Nuclear factor κB (NF-κB) signaling pathway plays a critical role in these biological processes, prompting us to investigate whether and how RAs affect this pathway. By screening compound libraries using NF-κB- dependent luciferase reporter, we observed that RA-V is the best NF-κB inhibitor. Further experiments demonstrated that RA-V interrupted the TAK1–TAB2 interaction and targeted TAK1 in this pathway. Moreover, RA-V prevented endotoxin shock and inhibited NF-κB activation and tumor growth in vivo. These findings clarify the mechanism of RA- V on NF-κB pathway and might account for the majority of known bioactivities of RA-V, which will help RA-V develop as new antiinflammatory and antitumour therapies.

Introduction

1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; To date, 56 RAs have been isolated from higher – Natural products derived from medicinal herbs have plants6 10, and these compounds have attracted con- been a rich source of leading compounds and have played siderable attention over the past three decades because of a vital role in drug discovery for centuries1,2. The roots their unique bicyclic structures and amazing antitumour and rhizomes of the Rubia plants, including Rubia cor- activities in vitro and in vivo, particularly RA-VII and RA- difolia, were recorded as a traditional Chinese medicine in V (Fig. 1a). Chinese Pharmacopeia and have been widely used for the Nuclear factor κB (NF-κB), a family of important tran- treatment of menoxenia, rheumatism, contusions, and scription factors, was discovered as a protein that binds tuberculosis. Rubiaceae-type cyclopeptides (RAs), qui- the immunoglobulin κ light chain gene enhancers in B nones, and triterpenes have been isolated from the Rubia cells11. The NF-κB family contains ﬁve proteins in – plants3 6. Among them, RA-VII, a Rubiaceae-type cyclo- mammals, including RelA (p65), RelB, c-Rel, NF-κB1, and peptide, was reported to have undergone phase I clinical NF-κB2, which are sequestered in the cytoplasm as het- trials at the NCI as an anticancer drug in Japan in 1990s3. ero- or homodimers. There are two distinct pathways: canonical and non-canonical pathways12. In the canonical NF-κB pathway the inhibitory protein IκBα is phos- Correspondence: Lijuan Chen ([email protected])or phorylated, ubiquitinated and degraded, which leads to Chen Wang ([email protected]) or Ninghua Tan ([email protected]) the nuclear translocation of the NF-κB complex and 1 School of Traditional Chinese Pharmacy & State Key Laboratory of Natural regulates the expression of the target genes12. Mounting Medicines, China Pharmaceutical University, Nanjing 211198, China 2State Key Laboratory of Phytochemistry and Plant Resources in West China, evidence has indicated that the NF-κB signaling pathway Kunming Institute of Botany, Chinese Academy of Sciences, Kunming 650201, plays a critical role in many biological processes and China controls the expression of over 500 target genes that are Full list of author information is available at the end of the article. These authors contributed equally: Zhe Wang, Simeng Zhao, Lihua Song Edited by Y. Shi

© The Author(s) 2018 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a linktotheCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 2 of 16

Fig. 1 (See legend on next page.)

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 3 of 16

(see ﬁgure on previous page) Fig. 1 RA-V inhibits TNF-α- and LPS-induced activation of NF-κB. a Chemical structure of RA-V. b RA-V inhibited the NF-κB signaling pathway in a dose-dependent manner. HEK293T or HeLa cells were transfected with the 5 × κB-luciferase and pTK-Renilla reporters. Twenty-four hours after transfection, the cells were incubated with various concentrations of RA-V for 6 h, and then treated with 10 ng/mL TNF-α for 2 h before the luciferase activity assay and MTT assay. c RA-V reduced TNF-α-induced expression of NF-κB target genes. HEK293T or HeLa cells were treated with various concentrations of RA-V for 24 h and stimulated with 10 ng/mL TNF-α for 2 h. The expression of the NF-κB target genes, IL-8, MCP-1, and E-selectin was measured by quantitative RT-PCR and normalized to GADPH expression. d RA-V inhibited TNF-α-induced p65 phosphorylation, IκBα phosphorylation and IκBα degradation. HEK293T or HeLa cells were incubated with various concentrations of RA-V for 24 h and treated with 10 ng/mL TNF-α for 10 min. The cell lysates were prepared and subjected to western blot analysis with the indicated antibodies. e RA-V inhibited TNF-α-induced IL-8 production. HeLa cells were treated with various concentrations of RA-V for 12 h before treatment with 10 ng/mL TNF-α for 2 h. The culture supernatant was collected and subjected to ELISA analysis. f RA-V inhibited the TNF-α-induced nuclear translocation of p65. HeLa cells were incubated with 200 nM RA-V for 6 h, treated with 10 ng/mL TNF-α for 15 min, and then subjected to immunocytochemical analysis. g RA-V reduced LPS-induced IL-6 expression. RAW264.7 cells were treated with various concentrations of RA-V for 24 h and stimulated with 1 μg/mL LPS for 3 h. The expression of IL-6 was measured. h RA-V inhibited LPS-induced p65 phosphorylation, IκBα phosphorylation. RAW264.7 cells were incubated with various concentrations of RA-V for 24 h and treated with 1 μg/mL LPS for 3 h. The protein expression was measured. i RA-V inhibited LPS-induced IL-6 production. RAW264.7 cells were treated with various concentrations of RA-V for 12 h before treatment with 1 μg/mL LPS for 3 h. The expression of IL- 6 was measured. The data in b, c, e, g and i are presented as the means ± S.D. from three independent experiments. *, p<0.05; **, p<0.01; ***, p< 0.001

involved in cell proliferation, apoptosis, angiogenesis, Notably, the potential roles for RAs in cancer therapy metastasis, and inflammation13,14. Abnormal activation of have been highlighted25. However, the underlying NF-κB pathway is frequently happened in many diseases, mechanisms and specific targets in the NF-κB pathway such as cancer, arthritis, and diabetes15,16. Thus, NF-κB remained unknown. We proposed that RA-V might exert has been an important therapeutic target, particularly in its effects by regulating NF-κB pathway. Therefore, we cancer, which has prompted significant effort to identify investigated how RA-V inhibits NF-κB activation in the moderators. At present, > 700 inhibitors of the NF-κB current study and identified that TAK1 is the RA-V target signaling pathway have been reported, but only a few have in this pathway. Moreover, RA-V powerfully blocks been developed into drugs14,17. For example, Bortezomib endotoxin shock and represses NF-κB activation and (Velcade), a reversible 26 S proteasome inhibitor, exerts tumor growth in vivo. These discoveries not only clarify its inhibitory activity on NF-κB at nanomolar concentra- the mechanism of RA-V in NF-κB pathway, but also tions and has been approved by the US FDA for the might account for the majority of known bioactivities of treatment of multiple myeloma18. RA-V, which will contribute to the future development of Transforming growth factor beta-activated kinase 1 RA-V as new therapeutic agents for the treatment of (TAK1/MAP3K7), an evolutionarily conserved member of cancer and inflammatory diseases. the mitogen-activated protein kinase kinase kinase family, is a key kinase upstream of NF-κB which can be activated Results by various stimuli and plays an essential role in this RA-V inhibits TNF-α- and LPS-induced activation of NF-κB pathway19. Accumulating studies have suggested an To discover small molecules that inhibit the NF-κB association between dysregulated expression of TAK1 and signaling pathway, we performed a cell-based screen using many human diseases, including cancer20. Therefore, small molecule libraries, including a library of RAs con- TAK1 is becoming an attractive therapeutic target for sisting of 29 natural and 5 synthetic RAs. We found that cancer treatment. RAs were one of potential natural NF-κB inhibitors, To identify potential natural NF-κB inhibitors, we particularly RA-V (Fig. 1a and Supplementary Figure 1), screened several small molecule libraries of ~ 200 com- and observed that RA-V dose-dependently inhibited pounds primarily derived from plants. Of the inhibitors TNF-α-induced activation of an NF-κB-dependent luci- that we identified, RA-V showed the best inhibitory effect ferase reporter in HEK293T and HeLa cells. In contrast, on NF-κB pathway. Our previous studies have demon- we observed only a slight antiproliferative effect from RA- strated that RA-V has antiinflammatory activity by inhi- V at high concentrations (Fig. 1b and Supplementary biting NO production and tumor necrosis factor (TNF)-α- Figure 2). To further confirm the inhibitory activity of induced NF-κB activation21; RA-V significantly suppresses RA-V on the NF-κB pathway, we investigated the effect of angiogenesis by downregulating ERK1/2 phosphorylation RA-V on NF-κB target genes by quantitative RT-PCR. As in HUVEC and HMEC-1 endothelial cells22; RA-V kills shown in Fig. 1c, RA-V dose-dependently decreased the human breast cancer cells by inducing mitochondria- TNF-α-induced expression of IL-8, MCP-1, and E- mediated apoptosis and inhibits cell adhesion and inva- selectin. Moreover, we examined the effect of RA-V on sion via the PI3K/AKT and NF-κB signaling pathways23,24. the expression of NF-κB-associated proteins. As expected,

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 4 of 16

Fig. 2 (See legend on next page.)

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 5 of 16

(see figure on previous page) Fig. 2 RA-V prevents endotoxic shock and inhibits NF-κB activation in vivo.a–c RA-V prevents endotoxic shock in vivo. Mice (n = 6 per group) were intravenously injected with RA-V micelles (0.25 mg/kg body weight) or control micelles 52, 28, and 4 h before an intraperitoneal injection of LPS (10 mg/kg). One hour later, the relative levels of IL-6 and TNF-α mRNAs in the liver were assessed by quantitative RT-PCR (a), the serum IL-6 and TNF-α levels were quantified by ELISA (b), and the expression of p65 and IκBα phosphorylation in the liver were measured by immunoblot analysis c. d RA-V improved animal survival. Mice (n = 20 per group) were intravenously injected with RA-V micelles (0.25 mg/kg body weight) or control micelles 52, 28, and 4 h before intraperitoneal injection of LPS (20 mg/kg). Animal survival was recorded in 2–3 hours intervals. e RA-V inhibits NF-κB activation in vivo. The NF-κB-luc transgenic mice (n = 4 per group) were intravenously injected with RA-V micelles (0.25 mg/kg body weight) or control micelles 52, 28, and 4 h before intraperitoneal injection with LPS (10 mg/kg). NF-κB-luc transgenic mice were photographed at different time points and representative images are shown. The bioluminescent signals from all of the mice were detected and quantified. The data are presented as the means ± S.D. *, p<0.05; **, p<0.01

western blot analysis demonstrated that RA-V also had the phosphorylation of IκBα and p65 was clearly inhibited dose-dependent inhibitory effects on TNF-α-induced in the liver tissue of RA-V-treated mice (Fig. 2c). In IκBα phosphorylation, IκBα degradation, and p65 phos- addition, RA-V also significantly improved the survival of phorylation (Fig. 1d). In addition, we found that RA-V mice treated with a lethal dose of LPS (20 mg/kg). As inhibited the TNF-α-induced production of IL-8 in a expected, after receiving the LPS injection, all of the dose-dependent manner using enzyme-linked immuno- control mice died within 36 h, whereas 60% of the RA-V- sorbent assay (ELISA) (Fig. 1e). The key step in canonical treated mice survived (Fig. 2d). To further confirm the NF-κB activation is the translocation of p65/p50 dimer in vivo effect of RA-V on the NF-κB pathway, we per- from the cytoplasm to the nucleus12, prompting us to formed the endotoxin shock model on NF-κB-luciferase perform an immunofluorescence assay. As showed in transgenic mice. The in vivo bioluminescence imaging Fig. 1f, RA-V significantly blocked TNF-α-induced p65 showed that RA-V inhibited NF-κB activation, particularly nuclear translocation. Besides, RA-V also inhibited lipo- 2 h after LPS injection (Fig. 2e). Collectively, these results polysaccharide (LPS)-induced interleukin (IL)-6 produc- demonstrated that RA-V prevents endotoxic shock in tion, IκBα phosphorylation and p65 phosphorylation in a mice and inhibits activation of the NF-κB signaling dose-dependent manner in RAW264.7 cells (Fig. 1g–i). pathway in vivo. Collectively, these results demonstrated that RA-V is a potent inhibitor of the NF-κB signaling pathway, with RA-V represses NF-κB activation upstream of the IKK IC50 values of 64.58 nM in HEK293T cells and 170.78 nM protein complex in HeLa cells, and that RA-V probably acts upstream of TNF-α and LPS have been shown to activate the p65 nuclear translocation. canonical NF-κB pathway through sequential interactions with TRAF2/MyD88, TAK1, IKK and p6515. To deter- RA-V prevents endotoxin shock and inhibits NF-κB mine where RA-V acts in this pathway, TRAF2, MyD88, activation in vivo IKKβ, or p65 plasmid was transiently transfected into The above results prompted us to explore the in vivo HEK293T cells together with the 5 × κB-lucieferase and effect of RA-V on the NF-κB signaling pathway. We pTK-Renilla reporters. RA-V dose-dependently sup- extended our study to the mouse endotoxin shock model, pressed expression of the NF-κB-regulated reporter in in which mice were intraperitoneally injected with LPS cells transfected with the TRAF2 or MyD88 plasmid, but (10 mg/kg or 20 mg/kg) and evaluated for changes by had no effect on expression in cells transfected with IKKβ examining physiological parameters induced by the or p65 (Fig. 3a). Western blots were performed to inves- inflammatory response26. tigate the effect of RA-V on NF-κB-associated proteins in First, the mice were intravenously injected with RA-V HEK293T cells transfected with the TRAF2, MyD88, or micelles (0.25 mg/kg body weight) or control micelles at IKKβ plasmid. As shown in Fig. 3b, RA-V dose-depen- three time points: 52, 28, and 4 h before they were dently inhibited p65 phosphorylation and IκBα phos- intraperitoneally injected with phosphate-buffered saline phorylation in cells overexpressing TRAF2 or MyD88, but (PBS) or LPS (10 mg/kg). One hour after LPS injection, we had no effect on expression in cells transfected with the examined proinflammatory cytokines (IL-6 and TNF-α) IKKβ plasmid. These findings indicated that RA-V might production at the transcriptional and translational levels. act upstream of the IKK protein complex. Moreover, we The expression of IL-6 and TNF-α mRNAs was sig- used a computational method to predict where RA-V is nificantly decreased in liver tissue of the RA-V-treated likely to act in this pathway. Using drugCIPHER27, a state- mice (Fig. 2a). The LPS- and RA-V-treated mice con- of-the-art network pharmacology tool that infers target sistently exhibited reduced expression of IL-6 and TNF-α profiles for drugs and small molecules in a genome-wide in the serum (Fig. 2b). Further analysis demonstrated that scale, and searching the STRING database28, we found

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 6 of 16

Fig. 3 (See legend on next page.)

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 7 of 16

(see ﬁgure on previous page) Fig. 3 RA-V represses NF-κB activation upstream of the IKK protein complex. a RA-V inhibited the TRAF2- and MyD88-induced expression of the NF-κB reporter in a dose-dependent manner. The indicated plasmids were transfected into HEK293T cells together with the 5 × κB-luciferase and pTK-Renilla reporters. Twenty-four hours after transfection, the cells were incubated with various concentrations of RA-V for 48 h before luciferase assays were performed. b RA-V inhibited TRAF2- and MyD88-induced p65 phosphorylation and IκBα phosphorylation in a dose-dependent manner. HEK293T cells were transfected with TRAF2, MyD88, IKKβ, or p65 for 24 h and then treated with various concentrations of RA-V for 48 h. The cell lysates were immunoblotted with the indicated antibodies. c RA-V likely exerts its inhibitory activity on the NF-κB pathway by acting on the TRAF6- TAK1–TAB1/2 complex. Using drugCIPHER, ARRB1 is a potential target of RA-V, as ARRB1 ranks 44th of 13388 genome-wide candidates. TRAF6- TAK1–TAB1/2 complex is a key link that connects ARRB1 and the NF-κB signaling pathway by searching the STRING database. The data in a and b are presented as the means ± S.D. from three independent experiments. *, p<0.05; **, p<0.01

Fig. 4 RA-V blocks the interaction between TAK1 and TAB2. a RA-V blocked the exogenous TAK1–TAB2 interaction. HEK293T cells were transfected with Flag-TAK1 and HA-TAB2 for 24 h and then incubated with various concentrations of RA-V for 6 h. The cell lysates were immunoprecipitated with HA antibody and then immunoblotted with the indicated antibodies. b, c RA-V does not disrupt TAK1–TAB1 or TAK1–TRAF6 interactions. HEK293T cells were transfected with Flag-TAK1 and HA-TRAF6 or HA-TAB1 for 24 h and then incubated with RA-V (800 nM) for 6 h. The cell lysates were immunoprecipitated with HA antibody, and immunoblots were performed with the indicated antibodies. d RA-V blocked the endogenous TAK1–TAB2 interaction. HEK293T (left) or RAW264.7 (right) cells were incubated with 800 nM RA-V for 6 h and stimulated with 10 ng/mL TNF-α for 2 h or 1 μg/mL LPS for 3 h, respectively. The cell lysates were immunoprecipitated with TAK1 antibody or control IgG and then immunoblotted with the indicated antibodies. Each experiment was repeated at least three times that RA-V likely exerted its inhibitory activity on NF-κB κB29,30. TNF receptor-associated factor 6 (TRAF6), an E3- signaling pathway by acting on TRAF6-TAK1–TAB1/2 ligase, is positioned upstream of TAK1 and recruits it in complex (Fig. 3c). Taken together, these results conﬁrmed response to various stimuli that induce NF-κB activa- that RA-V exerts its inhibitory activity on the NF-κB tion31. To explore the mechanism how RA-V inhibits NF- signaling pathway upstream of the IKK protein complex. κB activation upstream of the IKK protein complex, we analyzed the effects of RA-V on TAK1–TAB1, RA-V blocks the interaction between TAK1 and TAB2 TAK1–TAB2, and TAK1–TRAF6 interactions using co- TAK1, a serine/threonine kinase, is a member of the immunoprecipitation. We found that RA-V dose-depen- MAP3K family that regulates several signaling pathways, dently blocked the interaction between exogenously including the NF-κB signaling pathway11. In many dif- expressed TAK1 and TAB2 (Fig. 4a), but not between ferent cell types, TAK1-binding protein 1 (TAB1) and TAK1–TAB1 or TAK1–TRAF6 (Fig. 4b, c). To determine TAB2 form a stable complex with TAK1 in response to whether RA-V perturbed the interaction between endo- various stimuli and play a key role in the activation of NF- genous TAK1 and TAB2, we performed additional co-

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 8 of 16

Fig. 5 RA-V targets the TAK1 protein. a Chemical structure of CB12. b CB12 binds the exogenous TAK1. HEK293T cells were transfected with TAK1 for 24 h, then the cell lysates were incubated with biotin or CB12 followed by streptavidin agarose pull-down. The immunoprecipitates were incubated with a 10-fold excess of RA-V or biotin and western blotted with Flag antibody. c CB12 binds the endogenous TAK1. HEK293T cell lysates were incubated with CB12 or biotin and precipitated with streptavidin agarose, followed by western blotting with TAK1 antibody. d CB12 did not bind other key proteins of the NF-κB signaling pathway. HEK293T cells were transfected with the indicated plasmids for 24 h. The cell lysates were pulled down with streptavidin agarose and then immunoblotted with HA or Flag antibody. Each experiment was repeated at least three times immunoprecipitation experiments to examine the effect Because RA-V blocks the interaction between TAK1 of RA-V on the TNF-α- or LPS-induced interaction and TAB2, we explored whether TAK1 or TAB2 could between TAK1 and TAB2. As shown in Fig. 4d, RA-V bind to CB12. HEK293T cells were transfected with Flag- inhibited the interaction between TAK1 and TAB2. Col- TAK1 or HA-TAB2 for 24 h. The cell lysates were incu- lectively, these results revealed that RA-V blocks the bated with free biotin or CB12, and the mixtures were interaction between TAK1 and TAB2. precipitated with streptavidin agarose followed by western blot analysis. We found that Flag-TAK1, not HA-TAB2 RA-V targets the TAK1 protein was effectively pulled down by CB12 and the band could We attempted to synthesize chemical probes to identify be competed away by higher concentrations of unlabeled the potential RA-V targets responsible for its inhibitory RA-V (Fig. 5b), indicating that TAK1 likely binds to RA- effect on NF-κB activation. Synthetic derivatives were V. To further investigate whether RA-V binds to endo- designed based on the structure-activity relationship genous TAK1, we also performed a pull-down assay to between RAs and NF-κB (Supplementary Figure 1) and confirm the interaction between CB12 and TAK1. As the modifiable sites of the RA-V structure. We selected shown in Fig. 5c, the binding between CB12 and endo- the 2-alanine and 6-tyrosine amino acids as the mod- genous TAK1 was presented. Besides, no interaction was ification sites, and a series of analogs (CB1-12, Supple- found between CB12 and the other major NF-κB-related mentary Figure 3a) were synthesized, whose NF-κB proteins, including MyD88, TRAF2, TRAF6, TAB1, inhibitory activities were evaluated using the NF-κB- TAB2, IKKβ, and p65 (Fig. 5d). These results demon- dependent luciferase reporter. We found that NF-κB strated that TAK1 is the cellular RA-V target in the NF- inhibition was variably decreased for CB1-12, particularly κB signaling pathway. CB1 through CB5, which exhibited IC50 values > 20 µM. Fortunately, CB12 (Fig. 5a and Supplementary Figure 3b), RA-V binds the kinase domain of TAK1 a biotin-tagged RA-V, retained the ability to inhibit NF-κB To explore the mechanism in detail, we generated two signaling pathway with an IC50 value of 2.91 μM (Sup- Flag-tagged TAK1 truncation mutants, one of which plementary Figure 3c–d). Therefore, we chose CB12 as a encoded the kinase domain. We observed that RA-V only chemical probe to identify potential targets of RA-V. bound the kinase domain of TAK1 (Fig. 6a). Next, we

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 9 of 16

Fig. 6 RA-V binds the kinase domain of TAK1. a CB12 directly binds to the TAK1 kinase domain. HEK293T cells were transfected with TAK1-N or TAK1-C for 24 h. The cells lysates were incubated with CB12 or biotin and precipitated with streptavidin-coated sepharose and were analyzed by immunoblotting with Flag antibody. b The model for RA-V binding to the crystal structure of the TAK1–TAB1 fusion protein (PDB 4GS6). The blue dotted lines represent the hydrogen bonds between RA-V (green) and TAK1–TAB1. c RA-V probably binds the ATP-binding pocket of TAK1. HEK293T cells were transfected with TAK1 for 24 h, and then the cell lysates were incubated with biotin or CB12 before precipitation with streptavidin agarose. The immunoprecipitates were treated with ATP or (5Z)-7-oxozeaenol and western blotted with Flag antibody. d RA-V inhibits TAK1, JNK, ERK, p38, and IKKα/β phosphorylation. HeLa cells were incubated with various concentrations of RA-V for 12 h and then treated with 10 ng/mL TNF-α for 10 min. The cell lysates were prepared and western blotted with the indicated antibodies. Each experiment was repeated at least three times modeled the RA-V-TAK1 complex. The Auto Dock vina the proteins that TAK1 directly or indirectly phosphor- program was used to predict the binding site for RA-V on ylates, including IKKα/β, JNK, ERK, and p38 (Fig. 6d). the TAK1 protein (TAK1–TAB1 complex, PDB: 4GS6). A Collectively, these results suggested that RA-V binds the detailed analysis of the generated models showed that RA- kinase domain of TAK1, and the binding sites probably V might bind to the ATP-binding pocket of TAK1 (afﬁ- located in the ATP-binding pocket. nity free energy: 9.8 kcal/mol), and hydrogen bond interactions between RA-V and Asp-175, Asn-114, Ser-111, RA-V exerts antitumour effect by inhibiting NF-κB Val-42 were observed (Fig. 6b). An additional pharma- activation in vivo cophore screen using the pharmacophore generated by NF-κB signaling pathway could promote cell survival by the RA-V-TAK1 complex showed that the model gath- regulating target genes and always be activated aberrantly or ered RA molecules with NF-κB inhibitory activities, constitutively in many tumor cells14. Comparing the results indicating the accuracy of the predicted binding cavity in Fig. 1 with Supplementary Figure 2, we found that the (Supplementary Figure 4). Furthermore, we found that bioactivities against tumor cell proliferation and NF-κB (5Z)-7-Oxozeaenol, a potent ATP-competitive irrever- activation of RA-V have good positive correlation in vitro, sible inhibitor of TAK132,33, effectively competed with the which drove us to investigate whether RA-V suppresses interaction between CB12 and TAK1 (Fig. 6c) using a tumor growth in vivo by inhibiting NF-κB activation. pull-down assay. ATP was used as a control. We also The antitumour effects in vivo were evaluated in HCT116 observed that RA-V inhibited TAK1 phosphorylation and and HepG2 xenografts in nude mice. As shown in Fig. 7a, b,

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 10 of 16

Fig. 7 RA-V inhibits tumor growth and the expression of NF-κB target genes in a nude xenograft model. a–d Inhibition of the growth of HCT116 and HepG2 xenograft tumors by RA-V. Female athymic nude BALB/c mice bearing HCT116 (n = 8) or HepG2 (n = 7) xenograft tumors were intravenously injected with various concentrations of RA-V micelles or control micelles every other day. 5-FU (10 mg/kg) group as positive control. Effects of RA-V on the growth curves of subcutaneoue xenografts of HCT116 (a) and HepG2 (b) and effects on the tumor weight in the HCT116 (c) and HepG2 (d) models. e–f Tumors removed were photographed. g–h The expression of the NF-κB target genes, IL-8, MCP-1, and CXCL-1, in HCT116 (g) and HepG2 (h) tumor tissue from vehicle- and various concentrations of RA-V-treated group was determined by quantitative RT-PCR and normalized to GADPH expression. *, p<0.05; **, p<0.01; ***, p<0.001

treatment with RA-V signiﬁcantly reduced the growth rates tumor weight of mice treated with RA-V was also obviously of HCT116 and HepG2 tumor growth compared with the lower than the control (Fig. 7c–f). To assess the possible control by measuring tumor volume every other day. The toxicity of RA-V, body weight was measured every other

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 11 of 16

Fig. 8 Analysis of potential side effects for treatment of RA-V. a–b The change curves of body weights of BALB/c bearing HCT116 (a) or HepG2 (b) xenograft tumors (n = 8 or 7). c Representative hematoxylin–eosin staining of heart, kidney, spleen, lung, and liver from vehicle- and various concentrations of RA-V-treated group. d The evaluation of serum ALT, AST, and creatine kinase for vehicle- and various concentrations of the RA-V- treated group

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 12 of 16

day throughout the whole experiment. As shown in Fig. 8a, synthase37. Furthermore, RA-VII exhibits the ability to b, weight of mice in the RA-V-treated groups had no change the conformational structure of F-actin to induce obvious changes in comparison with the control. We also G2 arrest and exhibits antiangiogenic activity in vitro and investigated toxic pathological changes in main organs by in vivo38,39. Our previous studies have also shown that RA- hematoxylin–eosin staining. Microscopic examination V has antiproliferative, antiinflammatory and anti- – demonstrated that no obvious toxic pathological changes angiogenesis activities in many cell lines21 25.However, were found in heart, kidney, spleen, lung and liver in the the underlying mechanisms and RA targets in the NF-κB RA-V-treated groups compared with the control (Fig. 8c). pathway have remained unknown. Because the NF-κB The levels of serum ALT, AST and creatine kinase were signaling pathway is involved in many important biological also determined to assess the liver and heart toxicity, processes, such as cell proliferation, apoptosis, and tumor respectively. As expected, no significant changes were metastasis, we investigated the inhibitory mechanisms and observed (Fig. 8d). Furthermore, quantitative RT-PCR identified RA targets in the NF-κB pathway that might be analysis in tumor tissue from xenografts was performed to responsible for most of the known activities of RAs determine the mRNA levels of NF-κB target genes. As reported to date, including the anti-proliferative, antiin- showninFig.7g, h, the expression of IL-8, MCP-1, and flammatory, anti-angiogenesis, and apoptotic activities. CXCL-1 were remarkable decreased in the RA-V-treated Then, we focused on RA-V, the best one among the 34 groups compared with the control. Taken together, these RA derivatives, with an IC50 value of 64.58 nM, as assessed results suggested that RA-V exerts antitumour activity by the NF-κB-dependent luciferase reporter in in vivo without any discernible side effects by inhibiting NF- HEK293T cells. We confirmed that RA-V is a new natural κB activation. inhibitor using subsequent in vitro and in vivo experiments and came to several conclusions: (a) Cells treated Discussion with various concentrations of RA-V exhibit decreased According to the data from the World Health Organi- expression of TNF-α- and LPS-induced NF-κB-responsive zation (WHO), cancer, particularly of the lung, colon, and genes. (b) RA-V inhibits the expression of TNF-α- and breast cancer, is one of leading causes of death in many LPS-induced NF-κB-associated proteins in a dose- countries and is still increasing34. To reduce the mortality dependent manner. (c) The TNF-α-induced transloca- rates, many studies have focused on the occurrence and tion of the p65/p50 heterodimer is blocked by RA-V. (d) development of cancer to identify key targets or signaling In the LPS-induced endotoxic shock model, RA-V not pathways as potential treatment options. Among them, only reduces the expression of proinflammatory cytokines the NF-κB signaling pathway has drawn considerable and NF-κB-associated proteins but also consistently pro- attention owing to its constitutive activation in many longs survival time. (e) The inhibitory activity of RA-V cancers, including both hematopoietic and solid malig- was also confirmed in NF-κB-luciferase transgenic mice nancies35, prompting many researchers to develop NF-κB by bioluminescence imaging in vivo. Thus, these results inhibitors. Although > 700 compounds have inhibitory encouraged us to select RA-V as a representative com- effects on the NF-κB pathway, most of them block sig- pound for further analysis investigating inhibitory naling downstream of IKK, and only a few have been mechanisms and targets of the NF-κB signaling pathway. included in clinical trials or used as clinical cancer Although it is believed that the activities of the IKK remedies14. Therefore, the identification of novel NF-κB complex proteins and proteasome are important for NF- inhibitors with new targets is still needed. κB activation40,41, RA-V fails to inhibit their activities or To develop small molecules that inhibit the NF-κB sig- expression. Therefore, we explored where RA-V acts in naling pathway, we first performed a cell-based screen the NF-κB signaling pathway by biochemical and bioin- using small molecule libraries (~ 200 compounds) and formatics analyses. The results indicated that both found that RAs exhibited powerful inhibitory activities on TRAF2- and MyD88-induced NF-κB activation and the the NF-κB pathway. RAs that contain unique bicyclic expression of NF-κB-associated proteins are prevented by structures are potential natural antitumour compounds. RA-V, but not those induced by IKKβ and p65, indicating Among them, RA-VII was approved for phase I clinical that RA-V acts upstream of IKK complex. The bioinfor- trials by Japanese scientists in the 1990s, but its use was matics analysis also supports this interpretation. There- forcefully terminated owing to its toxicity and water solu- fore, RA-V exerts its inhibitory effect in a different way bility3. Nevertheless, much antitumour mechanistic inves- than most known NF-κB inhibitors. tigations and new drug study of RAs have been performed TAK1, a member of the MAP3K family that is a kinase during these years. Earlier studies indicated that RA-VII upstream of IKKβ in the NF-κB signaling pathway, has a suppresses protein synthesis by binding to eukaryotic 80 S critical role in the canonical pathway through its inter- ribosomes36. Moreover, RA-V and RA-XII significantly action with TAB1, TAB2, and TRAF629,30,42. RA-V inhibit NO overproduction and the inducible nitric oxide blocked both the endogenous and exogenous interaction

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 13 of 16

between both TAK1 and TAB2. Based on the structure- Beijing HFK Bioscience Co., Ltd. (Beijing, China). The activity relationship between RAs and NF-κB and the mice were maintained in specific pathogen-free condi- modifiable sites on the RA-V structure, we synthesized tions at Sichuan University. The animal experiments several different chemical probes to identify one (CB12) were performed in strict accordance with the regula- that was active. Using biotin-tagged RA-V (CB12) as a tions in the guide for the Care and Use of Laboratory probe, which has an IC50 value of the 2.91 μM, we dis- Animals of Sichuan University (IACUC number: covered that RA-V directly targets TAK1 and probably 20100318). binds to its ATP-binding pocket, excluding the direct binding to other important NF-κB-associated proteins Isolation and synthesis of RAs such as MyD88, TRAF2/6, and TAB1/2, which were also We isolated natural RAs from the Rubia plants R. supported by computational study. In agreement with yunnanensis, R. schumanniana, R. cordifolia, and R. – these results, RA-V inhibits the phosphorylation of pro- podantha7 9,47,48. The synthetic RAs were described in teins that are downstream of TAK1, such as IKKβ, JNK, our previous study49. ERK, and p3843,44. Unfortunately, several attempts at MS and crystallization of the RA-V-TAK1 complex failed. Synthesis and information about the chemical probes Taken together, our data suggest that RA-V blocks the The structures and methods for CB1-12 synthesis are interaction between TAK1 and TAB2 and directly binds described in Supplementary Figure 3 and the Supple- the kinase domain of TAK1. Importantly, to our knowl- mentary Notes. edge, this is the first report of a natural product that interrupts the formation of the TAK1–TAB2 complex in Cell lines and culture the NF-κB pathway. It is worth mentioning that RA-V HEK293T, HeLa, RAW264.7, HCT116, and HepG2 cell may inhibit the NF-κB pathway through various ways with lines were obtained from the American Type Culture different mechanisms, either inside or outside this path- Collection (ATCC) and were cultured in Dulbecco’s way, and we only focused on the NF-κB signaling modified Eagle’s medium (Invitrogen) containing 10% (v/ pathway itself in this study. In addition, this is the first v) heat-inactivated fetal bovine serum (Life Technologies) report to identify TAK1 as RA-V target, which might and 1% (v/v) penicillin–streptomycin (Invitrogen). The account for the majority of known bioactivities of RA-V. cells were maintained at 37°C in a humidified incubator Besides, the bioactivities against tumor cell proliferation with an atmosphere of 5% CO2/95% air (v/v). and NF-κB activation of RA-V have good positive correlation in vitro and in vivo, which indicated that Cell transfection and luciferase assay RA-V exerts antitumour effect by inhibiting NF-κB HEK293T or HeLa cells were seeded in 24-well plates activation. and transiently transfected with the 5 × κB-luciferase and An increasing number of studies have shown that TAK1 pTK-Renilla reporters using Lipofectamine 2000 (Invi- plays an essential role in many physiological processes, trogen) for 18 h. The cells were then incubated with dif- such as inflammation, tumourigenesis, and metastasis, ferent concentrations of the compounds for the indicated which makes it an attractive molecular target for the times and subsequently stimulated with 10 ng/mL TNF-α treatment of several human diseases20. Recently, for 2 h. The luciferase activity in the cell lysates was researchers found that TAK1 inhibition induces apoptosis analyzed using the Dual Luciferase Reporter Assay System in KRAS-dependent colon cancers and promotes NF-κB- (Promega). The luciferase reporter assays were performed dependent cell death in AML cells in vitro and in vivo45,46. as previously described50. Therefore, RA-V could be used to develop therapeutic agents that target TAK1. Overall, this study demonstrated MTT assay that RA-V is a potent new natural inhibitor of the NF-κB HEK293T cells (5 × 103 cells/well) or HeLa cells (3 × 103 signaling pathway, which explains its antiproliferative, cells/well) were seeded in a 96-well plate with 100 μL antiinflammatory, antiangiogenic and apoptotic effects. In medium and cultured for 12 h in a CO2 incubator. Then, future, we will focus on the role of RA-V in TAK1- the cells were treated with various concentrations of RA- associated biological activities or diseases, which will V or dimethyl sulfoxide (DMSO) for the indicated hopefully contribute to its further development as a new times before 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenylte- clinical remedy for cancer and other related diseases. trazolium bromide (MTT) solution (final concentration at 0.5 mg/mL) was added to each well. After 4 h, the medium Materials and methods was replaced with 150 μL DMSO, and the optical density Ethics statement was measured at 595/650 nM (Molecular Devices). Six- to eight-week-old BALB/c mice and female The results are presented as the cell growth inhibition athymic nude BALB/c mice were purchased from rate.

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 14 of 16

Quantitative RT-PCR with reverse transcription SuperSignal West Pico Chemiluminescent Substrate The total RNA was extracted from the indicated cells (Pierce). using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions and then was reverse tran- Pull-down of RA-V-bound proteins scribed. The RNA concentrations were measured spec- HEK293T cells were harvested and lysed in RIPA buffer trophotometrically with a BioPhotometer (Eppendorf). with protease inhibitors. After centrifugation at 12,000 Quantitative real time PCR for IL-8, MCP-1, E-selectin, rpm for 15 min, the supernatant was obtained and divided CXCL-1, IL-6, TNF-α,andGADPH was performed with into four equal parts, ensuring that the total protein the SYBR Green PCR Master Mix (ABI). The PCR pri- concentration in each sample was equivalent. The mers used to detect the target genes are listed in Sup- supernatants were precipitated with streptavidin agarose plementary Table 1. for 1 h before incubation with biotin or CB12 overnight at 4 °C. Then, the beads were incubated with DMSO, biotin Cytokine detection by ELISA or RA-V. After the incubation, the beads were washed Murine TNF-α and IL-6 were measured using ELISA four times with RIPA buffer, and the bead-bound proteins kits (R&D Systems) according to the manufacturer’s were eluted with 2 × SDS loading buffer. The samples specific instructions. were separated by SDS-PAGE, and the bound proteins were detected using the indicated antibodies. Immunofluorescence assay HeLa cells were plated at a density of 1 × 105 cells per Preparation of RA-V-loaded polymeric micelles glass coverslip for 24 h before RA-V treatment. After 6 h RA-V-loaded micelles were prepared using thin-film of incubation with RA-V, the cells were treated with 10 dispersion method. In brief, 10 mg RA-V and 200 mg ng/mL TNF-α for 20 min and fixed in 4% paraformalde- mPEG2000-PDLLA2000 were completely dissolved in hyde for 15 min. After permeabilization in 0.1% Triton X- 100 mL mixed solvent (80 mL methylene chloride and 20 100 for 20 min, the cells were incubated with 5% bovine mL methanol) to obtain a clear solution in a round- serum albumin for 1 h, followed by an overnight incuba- bottom flask. The solvent was removed by rotary eva- tion with an antibody to p65. After three washes in PBS, poration under reduced pressure at 60 °C for about 3 h the coverslips were incubated with a fluorescein and a thin layer of uniform film on the wall of the flask isothiocyanate-conjugated goat anti-mouse IgG (Jackson was formed. Saline was added to the flask and the mixture ImmunoResearch) for 1 h. The nuclei were counter- was stirred for 10 min at 60 °C to prepare a RA-V- stained with 4′,6-diamidino-2-phenylindole (Sigma- containing polymeric micelle aqueous solution. Then the Aldrich). The fluorescent signals were examined with a resultant solution was passed through 0.22 μm filter confocal microscope (Leica). membrane to remove the unincorporated RA-V aggre- gates. So the RA-V loaded micelles were obtained. Immunoprecipitation and western blot analysis The cell lysates were prepared using radio- Tumor xenograft in nude mice and toxicity assessment immunoprecipitation (RIPA) buffer containing 50 mM In total, 2 × 106 HCT116 or 3 × 106 HepG2 cells were Tris-HCl pH 7.4, 150 mM NaCl, 1 mM ethylenediamine- subcutaneously injected into the right dorsal flank of 6- to tetraacetic acid, 1% Triton X-100, 0.5% deoxycholate, and 8-week-old female athymic nude BALB/c mice. After 0.1% sodium dodecyl sulfate, to which protease inhibitors tumor volumes reached 100 mm3, the tumor-bearing (Roche) were added. The lysates were incubated with the mice were randomly divided into several groups. The indicated antibodies before adding 20 μL of Protein A/G mice bearing HCT116 (n = 8) or HepG2 xenograft agarose for 2 h. After four washes with the same buffer, tumors (n = 7) were received the intravenous injection of the immunoprecipitates were boiled in 2 × SDS loading vehicle micelles or various concentrations of RA-V buffer. micelles every other day. 5-FU (10 mg/kg) group as For western blot analysis, the samples were equally positive control. Body weights and tumor volumes were subjected to sodium dodecyl sulfate polyacrylamide gel recorded every other day. Tumor volumes were calculated electrophoresis (SDS-PAGE) and transferred to poly- using the following formula: tumor volume (mm3) = vinylidene difluoride membranes (Millipore). After 0.52 × length × width2. At the end of the experiments, the blocking with 5% nonfat milk in tris-buffered saline with mice were killed. Tumor xenografts of each mouse were tween 20, the membranes were incubated with the removed and weighed. Part of each tumor tissue was indicated antibodies for 2 h or overnightat4°C,followed stored at liquid nitrogen for quantitative RT-PCR analysis. by an incubation with a horseradish-peroxidase- Organs (heart, kidney, spleen, lung, and liver) from conjugated secondary antibody for 1 h at room tem- killed mice were also removed, fixed in the 4% for- perature. The protein bands were detected with the maldehyde solution for microscopic examination by

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 15 of 16

hematoxylin–eosin staining. Serum alanine transaminase (CCDC: 672443) was used as the ligand. The model of (ALT), aspartate transaminase(AST), and creatine kinase the RA-V-TAK1 complex was also used to generate the (CK) were determined using relevant kits (Nanjing Jian- pharmacophore for the pharmacophore screen using the cheng) according to the manufacturer’s specific Discovery Studio 4.0 package. instructions. DrugCIPHER Bioluminescence imaging This computational method was performed as pre- The bioluminescence imaging of the NF-κB-luciferase viously described27. transgenic mice was performed using a Xenogen IVIS Spectrum in vivo imaging system (Caliper Life Sciences). Statistical analysis The mice were anesthetized with isoflurane and intra- The Student’s t test was used for the statistical analysis. peritoneally injected with 150 mg/kg D-luciferin in PBS. A p value < 0.05 was considered statistically significant. After 15 min, the in vivo bioluminescence was measured The statistical analysis was performed using GraphPad at the indicated times. The bioluminescent signals from Prism software. all of the mice were detected and quantified. Acknowledgements Antibody information This work was supported by the National Natural Science Foundation of China (U1032602, 91013002, 31470428, 81225025, 81630103), the National Basic The antibodies used included the following: TAK1 Research Program of China (2013CB127505), the Fund of Chinese Academy of (5206, Cell Signaling Technology, dilution 1:2 000), IKKα Sciences (XDA09030301-4, Hundred Talents Program), the National New Drug β Innovation Major Project of China (2017ZX09309027), the Fund for (11930, Cell Signaling Technology, dilution 1:1 000), IKK Introduction of High-level Talents from China Pharmaceutical University, the (8943, Cell Signaling Technology, dilution 1:2 000), TAB2 Program for Jiangsu Province Innovative Research Team, the Fund Program of (3744, Cell Signaling Technology, dilution 1:1 000), IκBα State Key Laboratory of Natural Medicines (3144060028), and “111” Project (B16046) from the Ministry of Education of China and the State Administration (4812, Cell Signaling Technology, dilution 1:2 000), of Foreign Experts Affairs of China. We thank Dr. Yin ZN (Nankai University) for phospho-TAK1 (4508, Cell Signaling Technology, dilu- providing us with NF-κB-luciferase transgenic mice. We thank Li Y (Kunming tion 1:5 000), phospho-IKKα/β (16A6, Cell Signaling Institute of Botany, CAS) and Li J (Shanghai Institute of Materia Medica, CAS) for Technology, dilution 1:5 000), phospho-NF-κB p65 (3033, advices. Cell Signaling Technology, dilution 1:8 000), phospho- κ α Author details I B (2859, Cell Signaling Technology, dilution 1:2 000), 1School of Traditional Chinese Pharmacy & State Key Laboratory of Natural HA (sc-7392, Santa Cruz Biotechnology, dilution 1:2 000), Medicines, China Pharmaceutical University, Nanjing 211198, China. 2State Key and FLAG (F1804, Sigma, dilution 1: 5 000). Laboratory of Phytochemistry and Plant Resources in West China, Kunming Institute of Botany, Chinese Academy of Sciences, Kunming 650201, China. 3State Key Laboratory of Biotherapy/Collaborative Innovation Center of Plasmids Biotherapy and Cancer Center, West China Hospital of Sichuan University, 4 TAK1 cDNA was obtained from a thymus cDNA library Chengdu 610041, China. State Key Laboratory of Cell Biology, Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences, by PCR and subsequently inserted into the pcDNA3.1 Shanghai 200031, China. 5MOE Key Laboratory of Bioinformatics and vector. The TAK1 truncation mutants were generated Bioinformatics Division, TNLIST/Department of Automation, Tsinghua from TAK1 cDNA using PCR. The specific primers used University, Beijing 100084, China for PCR are listed in Supplementary Table 2. The reporter Conflict of interest plasmids (5 × κB-luciferase and pTK-Renilla) and the The authors declare that they have no conflict of interest. MyD88, TRAF2, TRAF6, TAB1, TAB2, IKKβ, and p65 plasmids have been described previously51. All of the fi Publisher's note plasmids were analytically veri ed by sequencing. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Purchased chemical compounds The chemicals and solvents used to synthesize the Supplementary Information accompanies this paper at (https://doi.org/ 10.1038/s41419-018-0743-2). derivatives and probes were purchased from J&K or Aladdin. mPEG2000-PDLLA2000 was obtained from Received: 31 July 2017 Revised: 21 May 2018 Accepted: 28 May 2018 Jinan Daigang Biomaterial Co.,Ltd.

Molecular modeling The Auto Dock vina docking program52 was used to References predict the RA-V binding site on the TAK1 protein. The 1. Koehn, F. E. & Carter, G. T. The evolving role of natural products in drug discovery. Nat. Rev. Drug Discov. 4,206–220 (2005). crystal structure of TAK1 (TAK1–TAB1 complex, PDB: 2. Newman,D.J.&Cragg,G.M.Naturalproducts as sources of new drugs over 4GS6) was used as the receptor, and the RA-V structure the 30 years from 1981 to 2014. J. Nat. Prod. 79,629–661 (2016). – modiﬁed from the crystal structure of RA-XXIII-Br 3. Tan, N. H. & Zhou, J. Plant cyclopeptides. Chem. Rev. 106,840 895 (2006).

Ofﬁcial journal of the Cell Death Differentiation Association Wang et al. Cell Death and Disease (2018) 9:715 Page 16 of 16

4. Zhao, S. M. et al. Antitumor cyclic hexapeptides from Rubia plants: history, 28. Franceschini, A. et al. STRINGv9. 1: protein-protein interaction networks, with chemistry, and mechanism (2005-2011). Chimia 65, 952–956 (2011). increased coverage and integration. Nucleic Acids Res. 41,D808–D815 (2013). 5. Xu, K. et al. Quinone derivatives from the genus Rubia and their bioactivities. 29. Shibuya, H. et al. TAB1: an activator of the TAK1 MAPKKK in TGF-β signal Chem. Biodivers. 11, 341–363 (2014). transduction. Science 272,1179–1182 (1996). 6. Xu, K. et al. Structural and bioactive studies of terpenes and cyclopeptides 30. Takaesu, G. et al. TAB2, a novel adaptor protein, mediates activation of TAK1 from the genus Rubia. Chem. Cent. J. 7, 81 (2013). MAPKKK by linking TAK1 to TRAF6 in the IL-1 signal transduction pathway. 7. Huang, M. B. et al. Rubischumanins A-C, new cytotoxic cyclopeptides from Mol. Cell 5,649–658 (2000). Rubia schumanniana. Tetrahedron 70,7627–7631 (2015). 31. Cao,Z.,Xiong,J.,Takeuchi,M.,Kurama,T.&Goeddel,D.V.TRAF6isasignal 8. Wang, Z. et al. Rubipodanin A, the first natural N-desmonomethyl Rubiaceae- transducer for interleukin-1. Nature 383,433–446 (1996). type cyclopeptide from Rubia podantha, indicating an important role of the 32. Ninomiya-Tsuji, J. et al. A resorcylic acid lactone, (5Z)-7-oxozeaenol, prevents N9-methyl group in the conformation and bioactivity. PLoS ONE 10, e144950 inflammation by inhibiting the catalytic activity of TAK1 MAPK kinase kinase. J. (2015). Biol. Chem. 278, 18485–18490 (2003). 9. Chen,X.Q.etal.RubicordinsA-C,newcyclopeptidesfromRubia cordifolia 33. Wu, J. et al. Mechanism and in vitro pharmacology of TAK1 inhibition by (5Z)- with cytotoxicity and inhibiting NF-kB signaling pathway. Tetrahedron 71, 7-oxozeaenol. ACS Chem. Biol. 8,643–650 (2013). 9673–9678 (2015). 34. World Health Organization (WHO) and International Union against Cancer 10. Figueiredo, P. O. et al. Rubiaceae-type cyclopeptides from Galianthe thalic- (UICC). Global Action against Cancer.(WHO/UICC,Geneva,Switzerland,2005). troides. J. Nat. Prod. 79, 1165–1169 (2016). 35. Baud, V. & Karin, M. Is NF-κB a good target for cancer therapy? Hopes and 11. Sen, R. & Baltimore, D. Multiple nuclear factors interact with the immu- pitfalls. Nat. Rev. Drug Discov. 8,33–40 (2009). noglobulin enhancer sequences. Cell 46,705–716 (1986). 36. Morita, H. et al. Conformational recognition of RA-XII by 80S ribosomes: a 12.Sethi,G.&Tergaonkar,V.PotentialpharmacologicalcontroloftheNF-κB differential line broadening study in 1HNMRspectroscopy.Chem. Pharm. Bull. pathway. Trends Pharmacol. Sci. 30, 313–321 (2009). 41,781–783 (1993). 13. Pahl, H. L. Activators and target genes of Rel/NF-kappa B transcription factors. 37. Tao,J.,Morikawa,T.,Ando,S.,Matsuda,H.&Yoshikawa,M.Bioactivecon- Oncogene 18, 6853–6866 (1999). stituents from Chinese natural medicines. XI. Inhibitors on NO production and 14. Gupta,S.C.,Sundaram,C.,Reuter,S.&Aggarwal,B.B.InhibitingNF-κBacti- degranulation in RBL-2H3 from Rubia yunnanensis: structures of rubianosides II, vation by small molecules as a therapeutic strategy. Biochim. Biophys. Acta III, and IV, rubianol, and rubianthraquinone. Chem.Pharm.Bull.51,654–662 1799,775–787 (2010). (2003). 15. Hayden, M. S. & Ghosh, S. Shared principles in NF-κB signaling. Cell 132, 38. Fujiwara, H., Saito, S. Y., Hitotsuyanagi, Y., Takeya, K. & Ohizumi, Y. RA-VII, a cyclic 344–362 (2008). depsipeptide, changes the conformational structure of actin to cause G2 arrest 16. Karin, M. Nuclear factor-κB in cancer development and progression. Nature by the inhibition of cytokinesis. Cancer Lett. 209,223–229 (2004). 441,431–436 (2006). 39. Koizumi, T. et al. Metronomic scheduling of a cyclic hexapeptide RA-VII for 17. Gilmore, T. & Herscovitch, M. Inhibitors of NF-κB signaling: 785 and counting. anti-angiogenesis, tumor vessel maturation and anti-tumor activity. Cancer Sci. Oncogene 25, 6887–6899 (2006). 97,665–674 (2006). 18. Mitsiades, N. et al. The proteasome inhibitor PS-341 potentiates sensitivity of 40. Scheidereit, C. IκB kinase complexes: gateways to NF-κB activation and tran- multiple myeloma cells to conventional chemotherapeutic agents: ther- scription. Oncogene 25, 6685–6705 (2006). apeutic applications. Blood 101, 2377–2380 (2003). 41. Skaug, B., Jiang, X. & Chen, Z. J. The role of ubiquitin in NF-κBregulatory 19. Yamaguchi, K. et al. Identification of a member of the MAPKKK family as a pathways. Annu. Rev. Biochem. 78,769–796 (2009). potential mediator of TGF-β signal transduction. Science 270, 2008–2011 42. Landström, M. The TAK1–TRAF6 signalling pathway. Int. J. Biochem. Cell Biol. (1995). 42,585–589 (2010). 20. Sakurai, H. Targeting of TAK1 in inflammatory disorders and cancer. Trends 43. Wang, C. et al. TAK1 is a ubiquitin-dependent kinase of MKK and IKK. Nature Pharmacol. Sci. 33,522–530 (2012). 412,346–351 (2001). 21. Fan, J. T. et al. Rubiyunnanins C-H, cytotoxic cyclic hexapeptides from Rubia 44. Johnson, G. L. & Lapadat, R. Mitogen-activated protein kinase pathways yunnanensis inhibiting nitric oxide production and NF-κB activation. Bioorg. mediated by ERK, JNK, and p38 protein kinases. Science 298,1911–1912 Med. Chem. 18, 8226–8234 (2010). (2002). 22. Yue, G. G. et al. Cyclopeptide RA-V inhibits angiogenesis by down-regulating 45. Singh, A. et al. TAK1 inhibition promotes apoptosis in KRAS-dependent colon ERK1/2 phosphorylation in HUVEC and HMEC-1 endothelial cells. Br.J.Pharm. cancers. Cell 148, 639–650 (2012). 164, 1883–1898 (2011). 46. Bosman, M. C. J. et al. The TAK1-NF-kB axis as therapeutic target for AML. Blood 23. Fang, X. Y. et al. Plant cyclopeptide RA-V kills human breast cancer cells by 124, 3130–3140 (2014). inducing mitochondria-mediated apoptosis through blocking PDK1/AKT 47. Fan,J.T.etal.RubiyunnaninsAand B, two novel cyclic hexapeptides from interaction. Toxicol. Appl. Pharmacol. 267,95–103 (2013). Rubia yunnanensis. Tetrahedron Lett. 51,6810–6813 (2010). 24. Leung, H. W. et al. Cyclopeptide RA-V inhibits cell adhesion and invasion in 48. Kuang,B.,Fan,J.T.,Zhao,S.M.&Tan,N.H.CyclopeptidesfromRubia schu- both estrogen receptor positive and negative breast cancer cells via PI3K/AKT manniana. Chin. J. Chin. Mater. Med. 37,2563–2570 (2012). and NF-κB signaling pathways. Biochim. Biophys. Acta 1853, 1827–1840 49. Liu, N. N. et al. Synthesis and conformation studies of rubiyunnanin B analogs. (2015). Tetrahedron 70, 6630–6640 (2014). 25. Lau,C.B.,Yue,G.G.,Fung,K.P.,Tan,N.H.&Leung,P.C.Thepotentialroleof 50. Zhao, S. M. et al. New cytotoxic naphthohydroquinone dimers from Rubia Chinese herbal medicines in cancer management. Science 347,S45–S47 alata. Org. Lett. 16,5576–5579 (2014). (2015). 51.Liu,X.,Chen,W.,Wang,Q.,Li,L.&Wang,C.NegativeregulationofTLR 26. Shi,M.etal.TRIM30α negatively regulates TLR-mediated NF-κB activation by inflammatory signaling by the SUMO-deconjugating enzyme SENP6. PLoS targeting TAB2 and TAB3 for degradation. Nat. Immunol. 9,369–377 Pathog. 9, e1003480 (2013). (2008). 52. Trott, O. & Olson, A. J. AutoDock Vina: improving the speed and accuracy of 27. Zhao, S. & Li, S. Network-based relating pharmacological and genomic spaces docking with a new scoring function, efficient optimization and multi- for drug target identification. PLoS ONE 5, e11764 (2010). threading. J. Comput. Chem. 31, 455–461 (2010).

Ofﬁcial journal of the Cell Death Differentiation Association